Error regarding implicit domain meshing using mmg3d

User Help
#1

Hi all,

I would like discretize implicitly defined cube domain in a rectangular computational domain.
I calculated .sol using Kratos and discretize using mmg3d as:
mmg3d_O3 silsoe_d90_m3_fluid -sol out_step=0.sol -ls -nr -hausd 0.1

and I get error *** Topological problem: non manifold surface at point 38602
Could you please tell me what the problem is?
The whole log is attached as follows.
The computational domain mesh (silsoe_d90_m3_fluid.mesh) can be downloaded from here: https://gigamove.rz.rwth-aachen.de/download/id/dcC2muCa6KgEGU
out_step=0.sol is this out_step=0.sol (1.1 MB)

Best regards,
Mayu

– MMG3D, Release 5.4.3 (Feb. 26, 2020)
Copyright © Bdx INP/CNRS/Inria/UPMC, 2004-
Jun 24 2020 15:16:04

– INPUT DATA
%% silsoe_d90_m3_fluid.mesh OPENED
%% out_step=0.sol OPENED
– DATA READING COMPLETED. 1.023s

&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&
MODULE MMG3D: 5.4.3 (Feb. 26, 2020)
&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&

– MMG3DLS: INPUT DATA
– INPUT DATA COMPLETED. 0.000s

– PHASE 1 : ISOSURFACE DISCRETIZATION

– MESH QUALITY 363004
BEST 1.000000 AVRG. 0.785829 WRST. 0.087757 (48247)

Warning: MMG5_chkBdryTria: 6 extra boundaries provided. Ignored

– PHASE 1 COMPLETED. 0.376s

– PHASE 2 : ANALYSIS
– PHASE 2 COMPLETED. 1.749s

– PHASE 3 : MESH IMPROVEMENT
*** Topological problem: non manifold surface at point 38602

Non orientable implicit surface. Exit program.

 NUMBER OF VERTICES      68837   CORNERS     1291
 NUMBER OF TETRAHEDRA   385805
 NUMBER OF EDGES           730   RIDGES       676
 NUMBER OF TRIANGLES     27705

– WRITING DATA FILE silsoe_d90_m3_fluid.o
%% silsoe_d90_m3_fluid.o.meshb OPENED
– WRITING COMPLETED

#2

Hello and welcome to this forum,

  • The first warning of Mmg (extra boundaries provided) means that some triangles are deleted by Mmg because they are not useful (it is not an issue because Mmg knows how to deal with it). In your case it is because they are in double (for example the triangles of indices 928 and 12404 are identicals)
  • The second issue is more complicated and creates the fatal error: your mesh contains overlapping elements (tetrahedra and triangles). You can look at the attached picture: for example triangles 922 and 12623 are overlapping.
    Overlap
    Overlap.png659×881 77 KB

Best Regards,
Algiane

#3

Hi Algiane,

Thank you for your reply. You were right about overlapping. I fixed it but still get an error.
The fixed .mesh file can be downloaded from here: https://gigamove.rz.rwth-aachen.de/download/id/cQhhDwUHXNTkAu
the .sol file is out_step=0.sol (911.9 KB)
I run the same command as before : mmg3d_O3 silsoe_d90_m3_fluid -sol out_step=0.sol -ls -nr -hausd 0.1
The log is attached as follows.

Best regards,
Mayu

– MMG3D, Release 5.4.3 (Feb. 26, 2020)
Copyright © Bdx INP/CNRS/Inria/UPMC, 2004-
Jun 24 2020 15:16:04

– INPUT DATA
%% silsoe_d90_m3_fluid.mesh OPENED
%% out_step=0.sol OPENED
– DATA READING COMPLETED. 0.920s

&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&
MODULE MMG3D: 5.4.3 (Feb. 26, 2020)
&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&

– MMG3DLS: INPUT DATA
– INPUT DATA COMPLETED. 0.000s

– PHASE 1 : ISOSURFACE DISCRETIZATION

– MESH QUALITY 308131
BEST 1.000000 AVRG. 0.786887 WRST. 0.109839 (76782)
– PHASE 1 COMPLETED. 0.315s

– PHASE 2 : ANALYSIS
– PHASE 2 COMPLETED. 0.397s

– PHASE 3 : MESH IMPROVEMENT
*** Topological problem: non manifold surface at point 36644

Non orientable implicit surface. Exit program.

 NUMBER OF VERTICES      56075   CORNERS        2
 NUMBER OF TETRAHEDRA   314377
 NUMBER OF EDGES            78   RIDGES        78
 NUMBER OF TRIANGLES     14320

– WRITING DATA FILE silsoe_d90_m3_fluid.o
%% silsoe_d90_m3_fluid.o.meshb OPENED
– WRITING COMPLETED

ELAPSED TIME 1.827s

#4

Hi Mayu,

I have pushed the correction of the bug in the master branch (in fact it was already corrected in the develop one).

However, even if the resulting mesh is valid, it as a poor quality (because the discretized isovalue is very close from the bounding box and almost coplanar with it).

If it is possible for you to use the develop branch of Mmg, it allows to use the -rmc option which removes small parasitic components within the level-set (Mmg find one of this components in your case). In your case it globally improves the quality of the mesh (even if the worst element remains very bad) and reduce slightly its final size.

I hope that it will be sufficient for you to run your test case!

Best,
Algiane